Thank you so much for information. I showed the code and the results of the training process, and noticed that the training accuracy is lower than the validation accuracy at the beginning of the experiments. Is there any problem if the training accuracy lower than validation accuracy???
Source: Deep Learning on Medium